ENHANCE - Environmental Change Institute€¦ · A particularly interesting aspect of flood management in England is the public-private partnership on flood insurance between the

ENHANCE Enhancing Risk Management Partnerships

for Catastrophic Natural Disasters in Europe

Grant Agreement number 308438

Deliverable 7.3: RISK SCENARIOS AND ANALYSIS

LONDON CASE STUDY: FLOOD RISK AND CLIMATE CHANGE

IMPLICATIONS FOR MSPs

Project 308438 • Risk scenarios and analysis: London case study ii

Title

RISK SCENARIOS AND ANALYSIS: LONDON CASE STUDY: FLOOD RISK

AND CLIMATE CHANGE IMPLICATIONS FOR MSPs

Author(s) Katie Jenkins, Jim Hall (UOXF); Swenja Surminski, Florence Crick (LSE)

Organization

UOXF, LSE

Deliverable Number

D 7.3

Submission date

26-02-2015

Prepared under contract from the European Commission Grant Agreement no. 308438 This publications reflects only the author’s views and that the European Union is not liable for any use that may be made of the information contained therein. Start of the project: 01/12/2012 Duration: 48 months Project coordinator organisation: IVM Due date of deliverable: Month 27 Actual submission date: Month 27 Dissemination level

PU Public

PP Restricted to other programme participants (including the Commission Services)

RE Restricted to a group specified by the consortium (including the Commission Services)

CO Confidential, only for members of the consortium (including the Commission Services)

Project 308438 • Risk scenarios and analysis: London case study iii

Executive summary

Flooding is recognised as one of the most common and costliest natural disasters in England and is

listed as a major risk on England's National Risk Register. Floods can take various forms, such as

river, coastal and surface water flooding. Surface water flooding represents flooding in urban areas

during heavy rainfall due to a combination of factors, including pluvial flooding (resulting from heavy

rainfall which does not infiltrate the ground but ponds or flows overland) flooding from sewers,

drains, and small watercourses. In the UK property is more likely to experience repeated surface water

flooding compared to fluvial or coastal flooding, costing England an estimated £1.3bn to £2.2bn per

year (Defra, 2011). In London surface water flooding is considered to be the most likely cause of

flood events, and probably the greatest short-term climate risk (Greater London Authority,

2009, 2011).

Additional to this threat is the potential impact of climate change on the frequency and intensity of

heavy precipitation events (IPCC, 2013). The UK Climate Projections (UKCP09) revealed that over

the course of this century, rainfall is projected to increase in winter but decrease in the summer,

although the number of days of heavy rainfall in summer will increase. Summers are therefore likely

to be characterised by intense heavy rainfall events intermixed with longer and relatively drier periods.

Winters, on the other hand will become wetter, with not only more average rainfall but also an

increase in extreme winter precipitation. These changes in winter and summer rainfall are expected to

lead to an increase in fluvial and surface water flooding (Ramsbottom et al, 2012).

In Greater London the number of residential properties in areas prone to surface water flooding has

been increasing from 2001 to 2011, as has the proportion of urban land covered with manmade

surfaces (which is >70% in many London boroughs) (HR Wallingford, 2012). Development in areas

prone to surface water flooding has been estimated at 0.5 to 0.7% per year from 2008 (Adaptation

Sub-Committee, 2012). These issues highlight that current land-use and development plans may

increase the exposure and vulnerability of society to surface water flood risk in Greater London.

Consequently, Defra estimate flood damage from surface water run-off could increase by 60-220%

over the next 50 years linked to changing precipitation patterns due to climate change, and

urbanisation (Adaptation Sub-Committee, 2012). The increased risk of surface water flooding and

need for further information has been highlighted as an area of concern within the London Plan (GLA,

2011), with effective and economically viable adaptation options required.

A particularly interesting aspect of flood management in England is the public-private partnership on

flood insurance between the UK government and the insurance industry known as the Statement of

Principles. Flood insurance is underwritten by the private market, while government commits to flood

risk management activities. However, the 2007 UK floods triggered a review of the Statement of

Principles. After more than two years of negotiation between government and industry, a new flood

insurance system Flood Re, was proposed by government in summer 2013, with the aim to finalise

and implement the new scheme by mid-2015. The proposed system, which creates an insurance pool

for properties at high risk of flooding, is presented by government and industry as a roadmap to future

affordability and availability of flood insurance, with an anticipated run-time of 20 to 25 years (Defra

and ABI 2013). The mechanisms of the new Flood Re scheme are still being negotiated, and to date

Project 308438 • Risk scenarios and analysis: London case study iv

there is little mention of how the MSP between government and insurers, and the new Flood Re

scheme could also promote effective flood risk reduction measures.

The London case study aims to provide an investigation of this public-private partnership and the

specific issue of surface water flood risk, and aims to analyse how the current and future proposals for

the partnership could influence London’s resilience to surface water flood risk today and in the future.

A key consideration is the incentives for risk reduction among different partners (including the

government, insurers, and developers), to support flood defences, household level flood protection,

and more appropriate spatial planning and zoning.

Crucial input into this investigation is a surface water flood risk assessment for Greater London. The

report presents a framework for evaluating current and future risk of surface water flooding in Greater

London. The study uses probabilistic hourly rainfall data from a spatially coherent version of the

UKCP09 Weather Generator (WG) (Kilsby et al., 2011) to identify heavy precipitation events for a

range of climate scenarios. A link is established to detailed Drain London1 surface water flood depth

maps, and damage to residential properties estimated using depth-damage functions.

The flood risk analysis highlights how this risk is expected to change in the future under projections of

climate change. Daily event frequency could increase by 54% by the 2030s and by 60% by the 2050s

compared to the baseline period. Average annual damage to residential properties is projected to

increase by 50% and 80% by the 2030s and 2050s respectively, suggesting an increase in severity of

events when they do occur. Based on the daily event data, the expected annual damage (EAD) is

calculated as £103, £184 and £198 million/year for the baseline, 2030 high and 2050 high climate

change scenarios respectively. An important component of the spatial WG is also the ability to provide

probabilistic results, with the range in results increasing from the baseline period to the future time

periods, representing the climate model uncertainty underlying the precipitation projections.

As a second step the event and damage data set is used as input within an agent-based model (ABM).

The model is parameterised based on large array of data sources and developed around GIS data to

allow a realistic representation of residential buildings and surface water flood risk. The ABM is

developed such that it can demonstrate the effect of flood risk and insurance on household wealth and

the potential for spatial shifts in inequality as a consequence of flood damage and insurance

(un)availability; assess the role of flood defences and PLPMs for risk reduction; and investigate the

existing public-private flood insurance partnership and the proposed new insurance scheme Flood Re.

The risk analysis provides a useful tool for decision making and risk management, and through the

proposed incorporation of this data in the ABM aims to support and justify future adaptation

strategies. These will be particularly important for Greater London where surface water flooding is

considered to be the most likely cause of flood events, and probably the greatest short-term climate

risk.

This will also be of great benefit to the MSP to highlight the potential benefits and limitations of the

scheme for risk reduction under future scenarios of climate change, broader economic consequences

for households and the housing market, potential benefits of investment in flood defences and

1 Led by the Greater London Authority. See: http://www.london.gov.uk/drain-london

Project 308438 • Risk scenarios and analysis: London case study v

household level protection in terms of reduced damages and reduced financial risk for insurers; and

the potential issues of future development in high risk areas for the scheme. Other components of the

scheme can also be tested, such as the inclusion of certain types of properties in Flood Re and the

financial implications of different transitional pathways to risk-based pricing of insurance in the

longer-term.

The above issues and questions are all highly relevant aspects for the ongoing regulatory and political

approval process for Flood Re, which have until now not received sufficient attention due to lack of

data or analysis. Our work addresses this gap and our findings are expected to provide important input

to the current discussion about the design and operation of Flood Re, particularly with regards to

incentivising flood risk reduction measures.

Project 308438 • Risk scenarios and analysis: London case study vi

Contents

1 Introduction ....................................................................................................................... 1

1.1 Surface water flood risk ........................................................................................................ 1

1.2 Flood risk management ......................................................................................................... 1

1.3 Aims and Objectives .............................................................................................................. 2

2 Specification of the risk analysis ...................................................................................... 4

2.1 Surface water flood risk in London ..................................................................................... 4

2.2 Modelling Framework........................................................................................................... 6

2.3 Spatial weather generator for urban areas ......................................................................... 7

2.4 Flood damage estimates ........................................................................................................ 8

2.5 Flood risk, insurance and adaptation strategies: An agent based model approach ........ 9 2.5.1 Modelling flood risk within the ABM ........................................................................... 12

2.5.2 Incorporating adaptation options ................................................................................... 12

2.5.3 Modelling the insurance market and Flood Re scheme ................................................. 14

2.5.4 Modelling the London housing market ......................................................................... 15

2.5.5 Modelling future development ...................................................................................... 16

2.5.6 Using the ABM to inform MSPs ................................................................................... 17

3 Results .............................................................................................................................. 18

3.1 Surface water flood risk analysis ....................................................................................... 18

4 Discussion of Results ....................................................................................................... 27

4.1 Surface water flood risk analysis ....................................................................................... 27

5 Conclusions ...................................................................................................................... 30

6 References ........................................................................................................................ 32

Project 308438 • Risk scenarios and analysis: London case study 1

1 Introduction

1.1 Surface water flood risk

Flooding is recognised as one of the most common and costliest natural disasters in England and is

listed as a major risk on England's National Risk Register. Floods can take various forms, such as

river, coastal, surface water, sewer and groundwater flooding, occurring independently but also in

conjunction, resulting in wide reaching impacts and consequences. Flooding affects not only people

and communities but buildings, agricultural land, infrastructure, ecosystems and often results in severe

displacement after an event. Flood risk management has been identified by the UK Climate Change

Risk Assessment as one of the high priority areas for action for the UK over the next five years, with

socio-economic factors including population growth, a changing climate, new development in the

floodplain and the relatively little understood surface water risk currently seen as the main challenges.

According to the UK CCRA coastal and surface flooding is the biggest potential impact of climate

change on the UK with the prospect for increasing losses and damage.

When comparing literature on fluvial or coastal flood risk, and surface water flood risk the latter is less

commonly investigated, and published articles specifically dealing with this type of risk are less

common. Correspondingly, there has been much less emphasis on conducting quantitative assessments

of economic and social costs of surface water flooding under future climate change, despite the

potential serious nature of impacts. At a broader level concerns have been raised that pluvial flood risk

across Europe may not be fully recognised, with some member states giving a much lower priority to

this type of risk (European Water Association, 2009).

In the UK property is more likely to experience repeated pluvial flooding compared to fluvial or

coastal flooding (Dawson et al., 2008), costing England an estimated £1.3bn to £2.2bn per year (Defra,

2011). Surface water flooding represents flooding in urban areas during heavy rainfall due to a

combination of factors, including pluvial flooding2, flooding from sewers, drains, and small

watercourses (European Water Association, 2009; Falconer et al., 2009). As such, heavy rainfall is

only one component in determining surface water flood risk. Drainage, often linked to sewage systems

in urban areas, and infiltration capacity related to land-use are also important factors. In the UK the

majority of surface water runoff drains directly into sewers, even from new developments.

Furthermore, as surface water flooding is often characterised by short duration and high intensity

rainfall events and so is difficult to predict, hindering the ability to forecast and warn communities of

flood risk.

1.2 Flood risk management

Flood management responsibility, policy and legislation for England are determined by the

Department for Environment, Food and Rural Affairs with national flood and coastal erosion

2 Pluvial flooding can be defined as flooding resulting from heavy rainfall which does not infiltrate the ground but ponds or

flows overland before the runoff enters a natural or man-made drainage system or watercourse, or where water cannot enter a

system as it is already at full capacity. Pluvial flooding is often characterised by short duration high intensity rainfall events

(typically over 20mm/h) (European Water Association, 2009).


management delivered by the Environment Agency. Local authorities have lead responsibility for

managing local flood risk, which includes surface water runoff, groundwater and ordinary

watercourses, and are designated as Lead Local Flood Authorities.

A particularly interesting aspect of flood management in England is the public-private partnership on

flood insurance between the UK government and the insurance industry known as the Statement of

Principles. Flood insurance across the United Kingdom is unique amongst most other national

schemes as it is purely underwritten by the private market, while government commits to flood risk

management activities. The Statement of Principles was established in 2000 in the wake of growing

flood losses and sets commitments from both the insurance industry and government to establish flood

insurance provision. The main obligations can be summarised as follows: flood insurance is provided

by private insurers under the Statement of Principles to both households and small businesses,

generally up to a risk level of 1:75 return period (RP) (1.3%) as part of their building and/or content

cover. Properties at higher risk are granted cover if insurers are informed by the Environment Agency

about plans for flood defence improvements for the particular area within the next five years.

Government commits to investment in flood defences and improved flood risk data provision as well

as a strengthened planning system. Under this agreement, the emphasis on flood risk reduction is

primarily placed on the government (national and local) as insurers play more of a financial supporting

role with little mention of how insurance can promote effective flood risk reduction measures.

The 2007 UK floods triggered a review of the Statement of Principles. After more than two years of

negotiation between government and industry, a new flood insurance system Flood Re, was proposed

by government in summer 2013. The Statement of Principles (SoP) officially ended on the 30th June

2013, but is still in operation whilst the political debate about the new Flood Re system continues,

with the aim to finalise and implement the new scheme by mid-2015. The proposed system, which

creates an insurance pool for properties at high risk of flooding, is presented by government and

industry as a roadmap to future affordability and availability of flood insurance, with an anticipated

run-time of 20 to 25 years (Defra and ABI 2013).

The Flood Re scheme will provide households under low to normal risk with standard insurance and

high risk properties will be insured through the Flood Re pool. The subsidy for high risk households is

claimed from a levy taken from all policyholders, on average £10.50 per policy, and also imposed on

insurers according to their market share. The premiums offered for high risk households are fixed

dependent on council tax banding and cover is offered at a set price based on what is felt to be initially

affordable. The government proposal is that small businesses will not be covered by the Pool unless

they operate from home with a domestic insurance policy in place. Policy excesses are intended to be

limited to between £250 and £500. Several other technical aspects remain unclear, including the

handling of flood losses beyond a suggested cap of 1 in 200 loss event, and will be subject to debate

between insurers and government.

1.3 Aims and Objectives

The main focus of the London case study is an investigation of the public-private partnership on flood

insurance between the UK government and the insurance industry. The case study aims to analyse


how this partnership could influence London’s resilience to surface water flood risk today and in the

future. A key consideration is the incentives for risk reduction among different partners, to support

flood defences, household level flood protection, and spatial planning and zoning, and how the

existing and proposed insurance schemes address this.

To achieve the objectives of the case study the analysis is conducted in two main parts:

i. The study uses probabilistic hourly rainfall data from a spatially coherent version of the

UKCP09 Weather Generator (WG) (Kilsby et al., 2011) to identify heavy precipitation events

for a range of climate scenarios. A link is established to detailed Drain London3 surface water

flood depth maps, and damage to residential properties estimated using depth-damage

functions. This facilitates the quantification of surface water flood risk in Greater London.

ii. Outputs from (i) are used within an agent-based model (ABM) developed for London. The

model is parameterised based on large array of data sources and developed around GIS data to

allow a realistic representation of residential buildings and surface water flood risk. The ABM

is developed such that it can demonstrate the effect of flood risk and insurance on household

wealth and the potential for spatial shifts in inequality as a consequence of flood damage and

insurance (un)availability; assess the role of flood defences and PLPMs for risk reduction; and

investigate the existing public-private flood insurance partnership and the proposed new

insurance scheme Flood Re.

The main focus of this report is related to part (i), outlining the analysis of future flood risk and

results. However, an introduction and overview of the ABM, its application to flood risk, adaptation

scenarios, and how it will be used to inform MSPs, is also provided.

3 Led by the Greater London Authority. See: http://www.london.gov.uk/drain-london


2 Specification of the risk analysis

2.1 Surface water flood risk in London

The study focuses upon Greater London (Figure 1), encompassing an area of 1,574 km2, a population

of approximately 7.5 million people, and 3.3 million residential dwellings (ONS, 2001).

Figure 1: The boundary of Greater London study area and location in England (inset)

London is considered well protected against tidal flooding but has a relatively low standard of

protection against surface water flooding. For example, surface water flooding in London in July 2007

affected over 400 properties, 158 schools and two hospitals (Environment Agency, 2010).

Consequently, in London surface water flooding is considered to be the most likely cause of flood

events, and probably the greatest short-term climate risk (Greater London Authority,

2009, 2011). More than 800,000 properties have been estimated to be at risk, and while most drainage

systems are designed to cope with a 1/30 year storm event, maintenance is often poor and so parts of

the network can perform below these standards (ibid.).

In Greater London the number of residential properties in areas prone to surface water flooding has

been increasing from 2001 to 2011, as has the proportion of urban land covered with manmade

surfaces (which is >70% in many London boroughs) (HR Wallingford, 2012). Over 96% of new

developments in London have been on brownfield sites, however, many of the remaining brownfield

sites for development lay in flood risk zones, with limited alternative sites for large scale development

(Greater London Authority, 2009). Similarly, development in areas prone to surface water flooding

has been estimated at 0.5 to 0.7% per year from 2008 (Adaptation Sub-Committee, 2012). These


issues highlight that current land-use and development plans may increase the exposure and

vulnerability of society to surface water flood risk in Greater London.

Additional to this threat is the potential impact of climate change on the frequency and intensity of

heavy precipitation events (IPCC, 2013). The UK Climate Projections published in 2009 (UKCP09)

revealed that over the course of this century, rainfall is projected to increase in winter but decrease in

the summer, although the number of days of heavy rainfall in summer will increase. Summers are

therefore likely to be characterised by intense heavy rainfall events intermixed with longer and

relatively drier periods. Winters, on the other hand will become wetter, with not only more average

rainfall but also an increase in extreme winter precipitation which is expected across the UK in all

regions. Projections for the 2080s under a medium emissions scenario suggest that in winter increases

in rainfall will vary between +10% and +30% over most of the UK, while in the summer there appears

to be a south to north gradient with a decrease in rainfall of close to 40% in south-west England to

almost no change in the north of Scotland ((Murphy et al., 2009)). These changes in winter and

summer rainfall are expected to lead to an increase in fluvial and surface water flooding ((Ramsbottom

et al., 2012)). Similarly, in England, Maraun et al., (2008) identified long-term increases in the

intensity of winter precipitation, and more recently increasing trends for spring and autumn. Fowler

and Ekström (2009) also projected increasing precipitation extremes in the UK for spring, autumn and

winter, increasing by 5-30% across different regions and seasons by 2070–2100.

Consequently, Defra estimate flood damage from surface water run-off could increase by 60-220%

over the next 50 years linked to changing precipitation patterns due to climate change, and

urbanisation (Adaptation Sub-Committee, 2012). The increased risk of surface water flooding and

need for further information has been highlighted as an area of concern within the London Plan (GLA,

2011), with effective and economically viable adaptation options required.

The Flood and Water Management Act 2010 gives London boroughs clearer responsibilities related to

surface water flood risk. New developments should utilise sustainable urban drainage systems (SUDS)

where possible, which aim to manage rainwater falling on roofs and other surfaces through a sequence

of actions. The main objective is to manage flow rate and volume of surface water runoff to reduce the

risk of flooding. Yet, the current uptake of SUDS is insufficient to mitigate increasing flood risk from

surface runoff, the risk of sewer overload, or to protect water quality (Defra, 2011). Incorporating

SUDS into new developments will only be part of the solution as new developments account for a

very small fraction of the housing stock. As such, retrofitting SUDS within existing urban areas and

developments will also be important for managing future flood risk (Environment Agency, 2007).

SUDS can also alleviate pressure on the sewerage network which will be crucial as in the UK the

majority of surface water runoff drains directly into sewers. Ellis and Viavattene (2014) estimate that

at least two-thirds of current urban flooding in the UK is the result of drainage system failure. Most

drainage systems in the UK are designed to cope with a 1/30 year storm event, but 50% of sewerage

systems are currently at or beyond capacity demanding major investment of an estimated £600m per

year (Defra, 2011).


Measures can also be implemented at a private level to reduce the severity of flood damage. For

example, resistance measures to keep water out such as door barriers, and resilience measure to reduce

the depth and damage caused by flood water like the use of water pumps and waterproof plaster. The

costs of PLPMs are reported to range from £3000 to £10,000 (Defra, 2008b; Harries, 2012), and have

been estimated to reduce flood damage costs by 65-84% (Thurston et al., 2008).

2.2 Modelling Framework

Risk based information on the potential economic and social impacts of surface water flooding, and

assessments of the effectiveness of adaptation measures, will be highly relevant if policy makers are to

address such concerns in the longer-term, and strive to implement more effective and efficient

adaptation measures. In developing such methodologies it is essential to recognise that surface water

flood risk and its associated impacts can be influenced by a variety of factors including the intensity

and spatial extent of the rainfall event itself; topography of the area; type of land use and infrastructure

in the region affected; permeability of the ground; surface and underground drainage capacity; and

interaction of pluvial flooding with other flood types such as fluvial flooding (European Water

Association, 2009). Furthermore, when focusing on complex and interrelated urban systems this

means that assessments of surface water flood risks and their impacts can span a wide range of

disciplines, government agencies, and actors.

Secondly, due to the characteristics of surface water flood events methodologies need to de developed

which can appropriately capture the localised scale and nature of the events, as well as the specific

characteristics of the urban areas affected. The scale and severity of economic and social impacts will

be dependent on the underlying vulnerability of the population and particular region exposed to the

event, as well as the underlying climate and weather patterns that determine the frequency and severity

of the event itself (Hall et al., 2005). This requires integrative thinking to understand and model

relationships between the urban environment, climate change, flood risk, and policy responses at a

local scale.

Fig. 2 and the below text provides an overview of the methodological framework (further details on

the main components are included in sections 2.3 – 2.4). The study uses flood depth maps for 1/30,

1/100, and 1/200 year return periods generated by the Drain London project. These maps reflect

present day climate, topographic data and information on the surface water drainage system at a fine

resolution (5 x 5m). Economic damages to residential building fabric and contents are calculated using

published flood depth-damage functions (Penning-Rowsell et al., 2010). To assess the potential impact

of climate change on future flood risk and economic damages, the corresponding return level of

extreme precipitation events of 1/30, 1/100, and 1/200 year return periods are estimated using hourly

rainfall data from a spatial version of the UKCP09 WG (Kilsby et al., 2011). This facilitates a

probabilistic analysis of extreme rainfall events, as well as providing an assessment of underlying

climate model uncertainties. Thus, for each flood event identified the spatial extent can be mapped, the

flood depth ascertained from the Drain London maps, and economic damage to residential buildings

estimated using depth-damage functions.


Figure 2: Overview of the risk based modelling framework which feeds into the ABM

2.3 Spatial weather generator for urban areas

The spatial and temporal scale of climate model outputs is often inconsistent with that required for

climate change impact studies. More spatially explicit climate projections can be produced by

incorporating downscaling techniques that account for local climatological features. The most recent

UK climate scenarios (UKCP09) have been accompanied by a stochastic Weather Generator (WG)

which can provide daily and hourly time series of weather variables for present and future conditions

at a 5 km2 resolution (Jones et al. 2009). The WG incorporates a stochastic rainfall model, which

simulates future rainfall sequences, and then generates other weather variables according firstly to the

rainfall state and then to other inter-variable relationships which are represented as regression

relationships. The WG has been well validated against observed data from 1961 to 1990 (ibid.).

The UKCP09 scenarios were novel in their representation of climate model uncertainties, based on the

range of climate model responses from a large perturbed physics ensemble (Murphy et al. 2009).

Results are presented as probability distributions of projected changes which can be used to

parameterise changes in the WG. In practice this is achieved by providing a Monte Carlo sample of

10,000 equiprobable vectors of change factors that are used to parameterise the WG. This leads to a

two-level sampling scheme in which (i) repeated representations of the WG for a given vector of input

parameters can be used to explore the effects of natural variability and (ii) sampling different vectors

of change factors explores the effect of climate model uncertainty in projected future impacts. In the

results reported here we refer to climate projections explored via (ii).

UKCP09 Projections

Weather Generator

Climate

Scenarios

Return level

(mm/hr) for return

periods (baseline)

Flood Risk Analysis

Building fabric &

content damage

dataset

Drain London flood

depth maps (1/30,

1/100, 1/ 200 year

return periods)

Database of extreme

rainfall events

Residential building

database (build type

/ age / location)

Depth-Damage

Functions

Direct economic

costs of surface

water flood events


The UKCP09 WG simulates weather sequences at a single site so does not provide spatial consistency

in time across neighbouring grid cells (Jones et al. 2009). The lack of spatial coherence limits the use

of the WG for analysing aggregate impacts over several grid cells. In this study a modified version of

the UKCP09 WG is used which provides spatially coherent time series data. Hourly precipitation

time-series data for 30 year stationary sequences (t) are taken from the WG for each grid cell (g) in the

study area. These series are generated 100 times each, with each run (r) based on a different randomly

sampled vector of change factors, to allow probabilistic analysis. This results in a 3-dimensional data

array of precipitation values A = [𝐴𝑔,𝑡,𝑛], where g = 71, t = 263,520, and r = 100. Data is generated for

the baseline period (1961–1990) and for the 2030s and 2050s under high emission scenarios

(equivalent to the IPCC SRES B1 and A1FI scenarios).

In order to assess surface water flood risk the return level of extreme precipitation events for 1/30,

1/100, and 1/200 year return periods is estimated from the baseline data. To calculate the recurrence

interval the hourly annual maximum series (AMS) is derived from the 100*30 year precipitation

times-series data for each grid cell. Extreme Value Analysis (EVA) is used to calculate the return

levels for each return period. The Generalised Extreme Value (GEV) distribution function is fitted to

the AMS. EVA allows the probability and return levels of extreme events to be determined, for return

periods exceeding that of the original data series, and even if events are more extreme than exists in

the data series (Sanderson, 2010). For each return period the GEV distribution allows the equivalent

return level to be determined for each grid cell. This gridded data provides the precipitation thresholds

above which surface water flooding of a given return period would be assumed to occur in each grid

cell. This is linked to damage costs of affected residential properties which intersect each grid cell in

the study area (outlined in section 2.4 below). These look up tables allow damages to be estimated for

each spatially explicit surface water flood event identified for the baseline and the 2030 and 2050 high

emission scenarios.

2.4 Flood damage estimates

Drain London provide surface water flood depth maps for Greater London based on precipitation data,

topographic data and information on the surface water drainage system. Flood risk is considered to

occur where flood depths exceed 0.1m as this is the depth where the significant onset of impacts, e.g.

damage to property, is considered to transpire (Penning-Rowsell et al., 2010). This study uses the UK

Buildings residential building class dataset4 which provides the spatial footprint of properties, property

type, and age for Greater London (It is assumed that there is no change in the number or spatial pattern

of residential buildings when assessing future scenarios). By overlaying the spatial flood maps onto

the building data it is possible to identify which properties are at risk of surface water flooding, and

the flood depth. As is the standard approach for flood loss analysis in the UK this study utilises the

damage functions provided in the Multi-Coloured Manual (MCM), for short (<12hr) duration floods

(Penning-Rowsell et al., 2010) (for more details on the application of depth damage functions also see

inter alia. Messner et al., (2007). Damage to residential building fabric and contents are estimated

4 The GeoInformation group data ® copyright by The GeoInformation® Group, 2014 Licence No. 3786.


based on building type (detached; semi-detached; terraced; flat), and building age (unknown; pre-

1919; 1919-1944; 1945-64; 1965-79; 1980-current).

Surface water flooding would not affect the whole of Greater London at the same moment in time. A

main benefit of the WG is that it is spatially coherent so an event affecting multiple grid cells can be

assessed in terms of damage for each property type and scenario, and results mapped to show the

spatial pattern of damages. In addition to assessing the economic damages from individual surface

water flood events, the average annual damage (EAD) can also be estimated based on the product of

the probability of a given flood event and damage integrated across all the scenario runs (e.g.

following Hall et al., (2006).

In order to summarise the flood event data and provide clear results, a daily event is defined as any

day when surface water flooding occurs in one or more grid cells for 1 hour or more. In the cases

where the same grid cell is affected for 2 or more consecutive hours the maximum hourly damage is

used in the estimates.

In addition, the spatial time series of flood events, and associated damages to residential properties,

provides the event set on which the effects of current and proposed public-private flood insurance

partnerships and incentives for adaptation will be tested within the ABM. In parallel with the

assessment of flood insurance, incentives for property level protection measures and flood defences

are also represented within the ABM.

2.5 Flood risk, insurance and adaptation strategies: An agent based model approach

ABMs are useful as they provide a bottom-up approach for understanding systems and their

behaviour, and are advantageous for visualising the effects of changing behaviours. To date ABM has

had limited application specifically to the insurance sector. Two papers have been identified which

look at consumer behaviour and insurance choices in general (Ulbinaitė et al., 2011; Ulbinaitė and Le

Moullec, 2010), while Brouwers and Boman (2011) focus more specifically on flood management

strategies, economic consequences, and decision making regarding flood insurance. More recently an

ABM has been developed which combines sea level rise and the impact of a flooding insurance

program, and models how these two aspects are likely to affect households’ location choices in a

coastal town in the US (Putra et al., 2014 submitted; Zhang et al., 2013). These studies establish a

bottom-up analytical framework to show how local real-estate markets respond to flood events and

how local governments choose adaptive actions.

In developing the ABM model code is adapted from the Putra (2014) model to capture the effects of

flood events and insurance on the local housing market. This code is adapted to the specific

characteristics of the UK and London, and extended to model insurance in a more elaborate fashion.

The ABM is parameterised based on large array of data sources and developed around GIS data to

allow a realistic representation of residential buildings and surface water flood risk. Fig. 3 provides an

overview of the model structure and design.


Figure 3: Overview of the agent based model

The ABM characterises five different agents: property owners, insurers, developers, government, and

bank agents, who interact within the environment. The key actions/features of the agents are

summarised as:

Property owners

Classed in the model as homeowners, buyers or sellers.

Decisions to sell are based on ability to pay mortgage and house fees, and the value of

property

Allocated incomes, housing fees, insurance costs

Homeowners are assumed to have flood insurance

People can also invest in Property Level Protection Measures (PLPMs)

- Classed as proactive and reactive responders to flood risk

Insurer

Main task to provide insurance to people

Sets premium and excess for insurance, based on risk

Allocate households to Flood Re scheme

Government

Can build/increase flood defences based on CBA

Give grants to people investing in PLPMs

The local government can sell land and evaluate development plans

Collect taxes from property owners and land sales, and grant from central government for

flood defences

Developer

React to the housing demand in the model

Locate land for building within opportunity areas


Submits development proposal to be approved by the local government.

Builds new houses and sells these houses on the market

Focus on making profit

Bank

A basic representation to allow the repossession of houses and selling these again on the

market

The properties and processes of these agents give rise to the aggregated results. The ABM runs in

annual time-steps and provides the user with options for spatial visualisation of key features (e.g.

house or person attributes (Figure 4)), switches to turn on/off different policies, flood events and

return periods, and graphs showing time-series of results. The following sections provide a description

of the main components, and agent behaviour, captured within the model.

Figure 4: Example of section of the user interface and visualisation: house view (build type)


2.5.1 Modelling flood risk within the ABM

Following the initial set up of the ABM the next key step is to impose a flood event on the study area.

Two approaches to modelling the flood ‘shock’ can be considered. Firstly, the model allows the user

to impose a flood event/s at set point/s in time via the user interface. This reads data from the daily

event dataset (from part (i) of the analysis) which provides information on the houses affected and the

damage to property fabric and contents for a given return period. Focusing on a few single events or

successive events is useful for model testing and validation. However, as this would represent one

single random event affecting the chosen area at any given time it is less robust for drawing policy

conclusions.

A second approach and area of planned future research is to use the probabilistic time-series data for

the analysis to test different policy options for the present day climate, and also considering future

effects of climate change. The daily event time series data is aggregated to annual steps (see Figure 6

below for an example). Where more than one daily event occurs in a single year the most damaging

event is used.

As well as incorporating data on the spatial pattern and level of damage for a given flood event an

estimate of flood risk is also required. This is calculated endogenously within the ABM to capture

dynamic changes in the model, and to allow risk to be assigned to new residential buildings created by

the model. Flood risk is calculated based on Bevan and Hall (2014, p.17). In any given year (t), the

risk (ri,t) is given by:

𝑟𝑖,𝑡 = ∫ 𝐷(𝑥𝑡)𝑓(𝑥𝑡)𝑑𝑥𝑡

∞

0

Where, 𝐷(𝑥𝑡) is a damage function with 𝑥 changing overtime, 𝑓(𝑥𝑡) is the flood probability

distribution. The probability of houses being affected by surface water flooding (for 1 in 30, 1 in 100,

and 1 in 200 year events, and damage data is available from part (i) of the analysis. As only three

points are available assumptions are made to extrapolating out from these points to create a damage

probability graph for each house. Namely, it is assumed that the function is linear between the three

known points. As the damage probability function will never have a probability of zero the function is

forced to meet the axis by assuming that the damage with zero probability is the maximum damage

that can be done to a house. There are many factors which would influence this at a household level,

and given a lack of data this is currently assumed to be 20% of the building value. The slope of the

function is assumed to extend horizontally until it crosses the x axis (damage). A maximum limit is

put on this assuming that it is highly unlikely that a house will be hit by a flood more than once every

2 years (i.e. 50% probability). Flood risk is then calculated as the area under the line.

2.5.2 Incorporating adaptation options

A key focus of the ABM is the investigation of the public-private partnership on flood insurance

between the UK government and the insurance industry, and how this partnership could influence

London’s resilience to major flood risk. This includes incentivising flood defences and household

level protection and as such options to adapt to flood risk are incorporated within the ABM. Alongside


the role of insurance and the Flood Re scheme PLPMs and SUDs, and government spending on flood

defences are incorporated. In the first case these options will be switched on and off in the model to

test them and their relative costs and benefits in isolation and combination (Table 1).

In the model it is assumed that people can be proactive and reactive in their behaviour to investing in

PLPMs. Harries (2008) reports that in 2004/05 6% of un-flooded households and 39% of previously

flooded households had taken steps to increase their resilience to flooding, and by 2008 the equivalent

figures remained almost unchanged at 9% and 34%, respectively. The value of 34% is used to reflect

the percentage of flooded households who will invest reactively in PLPMs in the model. However,

given that the flood protection association reported that in 2008 less than 5,000 homes had to date

taken approved measures (Defra, 2008a) a lower figure of 1% is used to represent the proportion of

proactive households in the model. The costs of PLPMs are estimated to range from £3000 - £10,000

(Defra, 2008b; Harries, 2012), with PLPLMs estimated to reduce flood damage costs by 65-84%

(Thurston et al., 2008).

Table 1: Scenarios to be tested in the ABM, including adaptation options

Current

insurance

system

Flood Re

system

Investment in

PLPMs

Investment in

flood defences

Off Off Off Off

ON Off Off Off

ON ON Off Off

Off Off ON Off

Off Off Off ON

ON ON ON Off

ON ON Off ON

ON ON ON ON

Within the ABM the local Government agent can approve or refuse applications for future

developments, and use their allocated flood protection budget to provide grants to houses to

incentivise PLPMs and build defences. A simple representation of the process of approving and

building flood defences is captured. The government is assumed to invest as much of its budget in

flood defence projects as possible, assuming that it can meet the cost from its budget and reflecting

that the project requires a cost benefit ratio of 1:5 or higher (Environment Agency, 2009). The local

government creates a flood defence portfolio of potential flood defence projects it can select to fund.

These projects will focus on the highest risk properties which have no existing defences, and capture

as many houses surrounding this while meeting the above approval criteria.

As the focus of the research is on surface water flood risk flood defences are represented by

investment in SUDs. There is limited information on the costs and benefits of SUDS, particularly for

urban retrofit, with data often location and case specific. When considering new developments Defra

(2011) assumed that SUDS could reduce flood damage by up to 35%. Overall, evidence suggests that

SUDS are cheaper to build (up to 30%) than traditional drainage systems, addressing the concern of


some developers that SUDS give rise to significant costs (Defra, 2011; Environment Agency, 2007).

This will be case specific, but even in complex cases costs are only assumed to be up to 5% higher

than traditional drainage system costs.

Defra (2011) report that on average 38% of minor developments and 58% of major developments are

now being built by developers with SUDS systems. It is not clear to what standards they are being

built to. It is assumed that 40% of build was accounted for by Minor Development with the remaining

60% accounted for by Major Development. This implies that 50% of build does not have SUDS.

Given the above data it is assumed that 50% of new builds created in the model will already have

SUDS at no extra cost. For retrofitting there is limited data and so it is assumed that the cost per house

to the government is £2000. In both cases risk will be reduced by 35% to estimate the potential

benefits of avoided damage. In future analysis we will also test the sensitivity of the above parameters

and options such as 1) that all new builds will include SUDS at no extra cost and reduce risk by 35%;

2) a current situation where 50% of new builds already have SUDS; and 3) a scenario where that there

is no uptake of SUDS and risk is increased by 35%.

2.5.3 Modelling the insurance market and Flood Re scheme

It is assumed that all home owners will have flood insurance initially, and will be required to pay an

insurance premium. In reality insurers can deny flood insurance to high risk properties, but in order to

gain insight into how the Flood Re scheme will address these high risk properties it is assumed

everyone can gain insurance, although at a given price. As the focus of the modelling is on the

technical rather than commercial risk the insurer will calculate flood risk based on the probability of a

given flood event and its associated damage (outlined in section 2.5.1), which will be used to calculate

the flood premium.

Home owners will also have an estimated insurance excess on their policy, which they will need to

cover in repairing damages following a flood event. It is assumed that the excess amount is non-

negotiable, and will be 1% (value based on expert opinion) of the insured value of the property.

Following any flood events the excess will also increase by 1/3rd

.

The insurer assets will reflect the income they receive in premiums and the compensation payments

they make following flood events, reflected in their loss ratio which will be calculated following each

flood event. Based on this value the insurer can increase or decrease prices accordingly.

Within the model there is also the option to ‘switch on’ the Flood Re scheme. The Flood Re scheme

will have assets in terms of a levy already paid as part of the insurance premium of households. If the

insurer decides the flood risk of a house is too high they are re-insured into the Flood Re scheme (if

built before 2009). The insurer will have to pay to re-insure the house into Flood Re, and income will

also be generated from the household premiums, capped based on the council tax band of the property

(where band A = £210; B = £210; C = $246; D = £276; E = £330; F = £408; G = $540; H = $540). In

this way the total compensation the insurer pays following a flood will be lower when the Flood Re

option is selected, as they are no longer required to compensate houses at highest risk.


2.5.4 Modelling the London housing market

As part of the London case study a particular area of interest is the implication of flood risk,

un/insurability, and the role of Flood Re on house prices and the housing market as a whole. There is

no current consensus on the impact of flood risk and insurability on property prices. Some empirical

studies have shown that property values drop immediately after a flood event, although they can

recover over time as memory of the event fades. For example, in a study of Albany, USA, Atreya and

Ferreira (2014) do not find a significant discount for floodplain properties before a flood event, but in

the wake of the 1994 flooding a significant discount is seen in property prices, which then diminishes

over time. Similarly, the UK based study of Lamond and Proverbs (2006) suggests that the impact of

flooding on house prices is temporary, lasting less than three years (Lamond and Proverbs, 2006). In

contrast, a UK study based on data on repeat sales found that high flood risk had no effect on property

values in areas with no recent flood events (Lamond et al., 2009), suggesting it is the actual flood

occurrence rather than perceived flood risk which has the largest effect on valuation of property prices.

To investigate these issues for the London case study the ABM has been developed to incorporate a

representation of the London housing market. The initial set up is based on the spatial location of

residential buildings from the UK Buildings database, with properties assigned data on property type,

age, council tax band, average property value per house type, level of flood risk, and the flood depth

and associated damage, if any, given a 1 in 30, 1 in 100, or 1 in 200yr flood event.

Consumers form the basis of the housing market with the ability to own, buy or sell houses. The initial

conceptualisation and design of the housing market is based on that of Zhang et al., (2013), who

propose a double auction market aimed at achieving high levels of market efficiency quickly by

running rounds of trading between buyers and sellers. Their model provides three choices: The buyer

can bid an amount for a property; the seller can ask an amount for a property; and there will be a

transaction price for the property. Where the bid and ask cross, the transaction price will be equal to

the earlier of the two. So, for example, a homebuyer forms a bid price between his income level,

adjusted by an affordability multiplier, and zero. A home seller randomly forms an ask price between

the maximum reasonable market price and his buying price of the property. A selected home buyer

then compares his bid with the best ask. If his bid is above the best ask, he accepts the best ask and the

transaction occurs at the best ask. If his bid is below the best ask (or there is no best ask) and there is

no best bid, it becomes the best bid. If his bid is below the best ask (or there is no best ask) and above

the best bid, it overrides the best bid. If his bid is below the best bid, the bid is ignored. The rule also

applies to a selected home seller. If his ask is below the best bid, a trade occurs at the best bid.

As in the model of Zhang et al., (2013) houses can be sold by consumers and the Bank agent, who can

repossess properties and sell these at a discounted rate on the market. In an extension of this the

London ABM also models future residential developments (see section 2.5.4 below), and as such

houses are also sold by the Developer. In running the model it is assumed that there are three separate

markets for the person, developer and bank to sell houses. It is assumed that consumers prefer newer

houses and so the developer market is run first, then the person market, and then the bank market. It is

assumed home buyers will also aim for the most expensive house and best possible house type first


and so houses are also sold based on this prioritisation from detached, semi-detached, and terraced,

down to flats.

Based on the modelled information on the number of houses on sale before the market run and houses

sold by the end of the run a ratio is created for each house type, and used to increase/decrease the

value of houses of this type accordingly. While house prices reflect the broader demand in the housing

market, consequences of flood events, insurability, and risk perception are incorporated as part of the

property owner’s decision making process when buying. Within the model buyers may or may not be

risk averse in their decision to consider past flood history and the flood risk level of a property. Risk

averse buyers can look back at a houses history, to a maximum of three years, to see whether it has

flooded (Lamond et al., 2009), and will not purchase a property if it has been flooded in the current

year. For houses flooded in the previous year 50% of risk averse buyers would not consider buying,

dropping to 25% for property flooded 2 years earlier. As such, there will be less demand for properties

where previous flood events have recently occurred.

In terms of selling decisions, property owners have three motives for selling their property in the

model. Firstly if they cannot afford the property fees i.e. annual fees such as the mortgage exceed their

income, in which case the bank can repossess their property. For flood affected home owners this will

also reflect the cost of insurance premiums, and the excess value they will have to cover following a

flood event. In the case of a property being uninsured this will reflect the total damage costs of the

flood which will need to be repaired. If a home owner cannot afford this additional cost then the house

is not repaired and the property value declines by this amount. Secondly a property owner may decide

to sell if they believe they can make a profit if the property value increases significantly above the

purchase value, and thirdly they can simply decide to move to a different house for other social factors

not considered explicitly within the model.

2.5.5 Modelling future development

As part of the housing market future building developments of residential properties are modelled.

This also allows for the role of developers and the local government to be investigated in terms of

testing different policy options and incentives for proposing and approving developments in high flood

risk areas.

In summary, the developer will establish the number of houses it wishes to build, based on the current

unmet demand for housing in the model. The developer will locate land for development based on

outlines of opportunity areas proposed for Greater London and the land value. Land value is calculated

based on the value of the surrounding houses. The developer will then create a development proposal

which will include the land it wishes to develop, the house type and the average value once built, and

the income cost ratio for the developer. If the income is higher than the expected cost then the

developer will submit this to the government for evaluation. In the model a development proposal will

be approved in 75% of the cases without a specific reason. Otherwise the development proposal will

be approved if the proposed flood risk of the development is lower than the maximum flood risk. If

this is also not the case the development proposal can still be approved if the development value at

risk (proposed building value if the house will be hit by a 1 in 200 year flood) divided by the


development income (proposed land value + proposed tax fee) is lower than the development approval

ratio.

All new properties in the model will be assigned their property type, a council tax band, flood risk

level, insurance premium and excess value, and the flood depth and associated damage, if any, given a

1 in 30, 1 in 100, or 1 in 200yr flood event.

2.5.6 Using the ABM to inform MSPs

The ABM has been developed in such a way that it is able to address multiple questions to help inform

the MSPs in London. Key questions which will be addressed include:

How Flood Re could be used to enhance resilience both in terms of physical risk reduction and

financial resilience. For example:

o how will design and coverage of the scheme effect risk (e.g. including/excluding new

build properties)

o modelling the different options for transition from Flood Re to risk based pricing

o How will present risk and risk under future climate change scenarios.

How insurance affordability could affect property prices and household decisions to buy/sell

in certain areas.

o Will `Ghetto' areas develop and if so where?

The implications of changing stakeholder/consumer information and knowledge on flood

awareness e.g. via communication, flood insurance, and government level adaptation

strategies

The effects of incentives to install flood resistance and/or resilience measures

o which agents could be best positioned to incentivise such measures e.g. government,

insurer, and developer?

How much could Property Level Protection measures (PLPMs) reduce flood risk/economic

damage?

What are the benefits of SUDs and other adaptation measures?

How could government affect flood risk

o i.e. government agents can refuse, permit or promote development in particular

locations?

The results of this research will contribute to the overall assessment of the existing public-private

flood insurance partnership and the proposed new insurance scheme Flood Re, and how this could

influence London’s resilience to flood risks today and in the future. This work is ongoing and a full

description of the method, sensitivity analysis, and results from part (ii) of the London case study will

be included within Deliverable 7.4.


3 Results

3.1 Surface water flood risk analysis

Extreme Value Analysis (EVA) was used to calculate the return levels of hourly precipitation, for each

return period, and for each of the 71 grid cells covering the Greater London study area. Results for the

grid cells were in the approximate range of 15-19.5mm, 17-23mm, and 18.5-25.5mm per hour for the

1 in 30, 100, and 200 year return periods respectively (Figure 5).

Figure 5: Return level of hourly precipitation (mm) for each return period across the 71 grid cells

The EVA provided the thresholds on which to estimate potential flood events of given return periods

and linked to flood depth and damage data. Using the Drain London flood depth maps the results

indicate that 312,551 to 392,758 properties could be at risk of surface water flooding (Table 2). This is

lower than the estimated 800,000 properties cited as being at risk from surface water flooding by the

GLA (2011). This may reflect the use of more accurate surface water flood depth maps from Drain

London, and the primary focus on residential properties only. Whilst there may also be some

limitations in the accuracy of the residential building database, the coverage was good with only

0.26% of total residential buildings being unclassified and as such excluded.

Table 2: Estimated economic impact of surface water flooding on residential buildings in Greater London

Return

period

Number of

properties affected

Damage to

building fabric

Damage to

building contents

Total damages

1/30yr 312,551 £2,319,104,560 £2,165,351,098 £4,484,455,658

1/100yr 352,942 £2,735,821,529 £2,498,683,701 £5,234,505,230

1/200yr 392,758 £3,140,705,414 £2,825,877,579 £5,966,582,992

Assuming a flood event affected the whole of Greater London results in total economic damages of

approximately £4.48bn to £5.97bn for 1/30 and 1/200yr flood events (Table 2), highlighting the


potentially severe economic implications of large scale events in London. As a comparison flood

claims following the UK 2007 floods (including surface water flooding) were estimated to be £3.2bn.

Figure 6 provides a breakdown of economic costs per property type and return period. Whilst the

largest proportion of residential building type in London is flats it should be noted that the building

dataset represents the spatial footprints of residential buildings. In the case of flats this will not

represent the upper storeys and individual flat units. However, this is not considered an issue for the

analysis as it is assumed that only the ground floor properties would be affected by surface water

flooding.

Figure 6: Breakdown of total economic damage by property type and return period

Figure 7 presents maps of flood risk, aggregated at a ward level for Greater London. The maps

highlight the spatial pattern of risk based on the flood extent and depth data, but also represents the

concentration and number of residential buildings in each ward, and the type of properties affected.

The figure highlights particular risk hotspots in Greater London, particularly in the South and South-

East.


Figure 7: Spatial maps of total economic damage to residential properties from surface water flooding for a) 1 in 30yr return

period, b) 1 in 100yr return period, and c) 1 in 200yr return period

However, surface water flooding would not affect the whole of Greater London at the same moment in

time. A benefit of using the probabilistic spatial WG is that long time series of daily events can be

provided for each grid cell (and aggregated to annual time-steps for use in the ABM as discussed

above), and as the WG is spatially coherent events affecting multiple grid cells can also be assessed,

with results reflecting both the intensity and spatial patterns of events across Greater London. Figure 8

presents a snap shot of a 100 year time slice showing the annual occurrence of flood events of

different probabilities for three grid cells in Greater London, for the baseline (top) and 2050 high

scenario (bottom). The graphs illustrate how areas of Greater London may be affected at different

times or simultaneously by surface water flood events, and how climate change is likely to increase

the severity and frequency of events in the future with the potential for areas to be affected in

successive years. Figure 9 illustrates the spatial extent and severity of two different rainfall events

identified.


Figure 8: Annual time-series of flood events for three individual grid cells in Greater London

Figure 9: Spatial maps showing two examples of extreme precipitation events and severity


Table 3 presents a summary of the estimated daily event frequency, average number of grid cells

affected and average damage per event and per year. The results are estimated based on the entire

hourly data array and presented for daily events which fall within the thresholds of individual return

periods.

The average values shown reflect the wider spatial extent of 1 in 30yr events, resulting in larger costs

per event than for 1 in 100 or 1 in 200yr events (although it should also be noted that average values

also mask large variation in flood losses of individual events). For example 1 in 30yr events affected

an average of 8 grid cells under the baseline scenario, compared to 5 grid cells for 1 in 200yr events.

Similarly, even though damages during 1 in 100 and 1 in 200 year flood events have the potential to

be much higher, the increased frequency of events of 1 in 30 years means that their contribution to

overall flood risk is very large (Table 3 and Figure 10).

Figure 10: Present and future flood risk (million £/yr) for Greater London


Table 3: Frequency of surface water flood events and associated damages

Baseline 2030 High 2050 High

Return

Period

Total

number

of daily

events

Average

damage per

daily event

(£ 2010)

Average

annual

costs (£

2010)

Average

number

of grid

Cells

affected

per

daily

event

Total

number

of daily

events

Average

damage per

daily event

(£ 2010)

Average

annual

costs (£

2010)

Average

number

of grid

Cells

affected

per

daily

event

Total

number

of daily

events

Average

damage per

daily event

(£ 2010)

Average

annual

costs (£

2010)

Average

number

of grid

Cells

affected

per

daily

event

1/30yr 1329 489,944,906 593,021 8 2081 440,847,688 996,123 7 2193 529,741,051 1,058,035 9

1/100yr 434 384,744,087 152,076 5 806 386,906,864 284,014 6 855 397,030,988 309,164 6

1/200yr 275 404,253,759 101,248 5 562 442,284,054 226,379 6 609 463,169,045 256,894 6

All

RPs 1588 535,835,292 774,960 7 2450 522,359,015 1,165,754 7 2537 602,463,466 1,392,031 8


The analysis highlights that overall event frequency could increase by 54% by the 2030s and by 60%

by the 2050s compared to the baseline period, although the average area of events remained largely

unchanged. The relatively small change in frequency of events between the 2030s and 2050s is in line

with other studies. For example, the study by Sanderson (2010) for London used UKCP09 data to

illustrate that rainfall events are likely to become more frequent in the future, particularly between the

present and 2040s. This could be because extreme rainfall events, particularly for longer return

periods, are already near to the maximum possible return levels. Hence, there will be a maximum

amount of precipitable water in the atmosphere and whilst the atmosphere can hold more water as it

warms, the incremental change in these extremes could become progressively smaller as the

atmosphere warms.

However, average annual damages were projected to increase by 50% and 80% by the 2030s and

2050s respectively, suggesting an increase in severity of events when they do occur. Based on the

daily event data, the expected annual damage (EAD) is calculated as £103, £184 and £198

million/year for the baseline, 2030 high and 2050 high climate change scenarios respectively.

An important component of the spatial WG is also the ability to provide probabilistic results. Table 4

presents results for all surface water flood events (i.e. where events are equal to and exceed the 1/30yr

return threshold), for the baseline and 2030 high and 2050 high climate scenarios, and also

disaggregated across the 100 model runs to provide an indication of the model uncertainty. Results are

presented for the median (50th percentile) and 10

th and 90

th percentile range.


Table 4: Percentile results showing the frequency of surface water flood events and associated damages for All RPs

Baseline 2030 High 2050 High

frequency

average cost

per year

average cost

per event frequency

average cost

per year

average cost

per event frequency

average cost

per year

average cost

per event

10th 13 90,549,441 175,145,317 14 155,401,104 244,841,515 13 180,389,736 306,969,764

50th 17 289,261,556 526,178,194 28 457,602,303 498,244,867 25 562,214,261 583,049,021

90th 23 585,393,662 914,098,752 57 1,117,925,503 853,175,802 48 1,172,357,582 1,073,215,947


The results suggest that the frequency of events will increase by the 2030s and 2050s compared to the

baseline, however, with a larger increase by the 2030s compared to the 2050s. In contrast the average

cost of surface water flood events per year increases from approximately £289 million per year (50th

percentile) in the baseline, to £458 million and £562 million per year in the 2030s and 2050s

respectively. As indicated above this suggests that while the frequency of events may increase in the

shorter term this may include more smaller scale events, compared to the longer term results which

have less events but when they do occur they are of a larger spatial extent and severity. This is also

highlighted when looking at the average cost per event. This declines from approximately £526

million per event (50th percentile) in the baseline to £498 million in the 2030 high scenario, but

increases to £583 million under the 2050 high scenario

Figures 11 and 12 present these results, along with the 10th and 90

th percentile range. In both examples

the range in results increases from the baseline period to the future time periods, representing the

climate model uncertainty underlying the precipitation projections.

Figure 11: Cost of surface water flood events per year. Results are presented for the median and 10th and 90th percentile

range

Figure 12: Frequency of surface water flood events per year. Results are presented for the median and 10th and 90th

percentile range


4 Discussion of Results

4.1 Surface water flood risk analysis

The main aim of the above work was to establish a link between outputs from a probabilistic urban

spatial WG and the detailed flood depth maps from Drain London to evaluate current and future risk

of surface water flooding in Greater London. Comparing literature on fluvial or coastal flood risk and

surface water flood risk highlights that the latter has received much less attention. Concerns have been

raised that pluvial flood risk across Europe may not be fully recognised, with some member states

giving a much lower priority to this type of risk (European Water Association, 2009). In the UK the

recent Climate Change Risk Assessment, which assessed changing flood risk under future climate

change, also focused on tidal and fluvial flooding, whilst implications of climate change for pluvial

and surface water flooding were not considered (although evidence from studies conducted by e.g. the

Environment Agency, Defra, and the GLA were considered) (Adaptation Sub-Committee, 2012).

Correspondingly, there has been less emphasis on quantitative assessments of economic costs of

surface water flooding for the present day and under future climate change, with a need to improve

methods and models for loss estimation. As such the above methodological framework applied to

Greater London provides a step to address this current gap in research.

It was estimated that 312,551 and 392,758 residential properties would be at risk of surface water

flooding following a 1 in 30 and 1 in 200 year event in Greater London. This was lower than the

estimated 800,000 properties cited as being at risk by the GLA (2011). However, it was postulated that

this could reflect the use of a different methodological approach, including the more recent and

detailed Drain London depth maps, and use of the UK Buildings dataset which carries its own

limitations. As further information on the method underlying the figure cited by the GLA could not be

found it is hard to provide a further comparison and fully verify this difference at this stage. Whilst

results may be considered conservative the economic damages estimated were still significant, ranging

from £4.4bn to £5.97bn for 1 in 30 and 1 in 200 year events across Greater London. The spatial

analysis also allowed risk hot spots to be identified (Figure 4), namely wards in the south and south-

east, where flood damage could reach approximately £50 million.

Given the localised nature of surface water flood events the analysis also allowed specific daily events

to be identified. Based on this event dataset the analysis highlights that overall event frequency could

increase by 54% by the 2030s and by 60% by the 2050s compared to the baseline period, although the

average area of events remained largely unchanged. Average annual damages were projected to

increase by 50% and 80% by the 2030s and 2050s respectively, suggesting an increase in severity of

events when they do occur. In comparison, Defra estimate flood damage from surface water run-off

could increase by 60-220% over the next 50 years linked to changing precipitation patterns due to

climate change, and urbanisation (Adaptation Sub-Committee, 2012). Again, results of this study

appear to be conservative, falling with in the lower end of this estimate. This discrepancy may be due

to the inclusion of other damage costs within the Defra estimates, but also that this study does not

consider additional effects of urbanisation such as new developments, especially in flood areas of high

flood risk, and increasing manmade surfaces and reduced greenspace.


The probabilistic analysis also highlighted how annual damages were projected to increase, although

the frequency of events was higher under the 2030 high scenario compared to the 2050 high scenario,

suggesting an increase in severity of events when they do occur. Results were presented for the 50th,

10th and 90

th percentile range and indicated the uncertainty in results due to underlying uncertainty in

the climate model projections. This uncertainty increases for both the 2030 and 2050 high scenarios

compared to the baseline.

Based on the daily event data, the expected annual damage (EAD) is calculated as £103, £184 and

£198 million/year for the baseline, 2030 high and 2050 high climate change scenarios respectively.

Assuming all else remains static this highlights both the economic impact that climate change could

have, but also the potential cost effectiveness of benefits that could be gained through investment in

adaptation options such as property level protection measures, SUDs, and urban greening. The role of

adaptation for risk reduction will be investigated further in the next stage of the analysis via the ABM

(see section 2.5.2). Additionally the methodology presented here could also be used to assess benefits

of adaptation. While there is limited data on the specific benefits of such measures on resultant flood

depth and damage that could be applied to Greater London as a whole, one option would be to conduct

a sensitivity analysis of the perceived benefits of adaptation. For example, assuming an upgrade to the

drainage system in which it could cope with a 1 in 100 year event rather than the current 1 in 30 year

event.

Whilst the methodology provides a novel approach to estimating surface water flood risk for a large

area there are also issues and limitations of any such study. In particular some important points to be

noted are:

Flood management is increasingly adopting a risk based approach, often using a range of

flood risk probabilities to provide a range of costs. However, as noted by Ward et al., (2011)

there is little insight into how these return periods are selected, how many should be

considered, and their impact on the outcomes of the study. This study adopted three return

periods in line with the available Drain London flood depth maps and the methodological

approach of the GLA.

Whilst the study benefits from the use of the spatial WG, which considers geographic and

topographic variations, other important processes which can affect rainfall, and which occur

over small spatial scales, are excluded and could result in model uncertainty.

Although considerable effort has been made to improve the rainfall model care should be

taken in using the WG to reproduce and analyse extreme precipitation events. The user is

advised to carry out uncertainty analysis based on the WG runs (Jones et al., 2009), and it is

noted that uncertainty will increase for longer return periods. The version of the WG used in

this study allows the effect of climate model uncertainty in projected future impacts to be

explored as discussed above.

As noted the probabilistic spatial WG allows uncertainty related to the physical climate

properties to be investigated. However, uncertainty surrounding the economic damages is

harder to ascertain. Flood depth-damage functions have limitations due to uncertainty in the

underlying data and assumptions which preclude their creation, and the influence of other

factors than depth on flood damage, such as velocity, duration, rise rate, and time of

occurrence (Messner et al., 2007). Even so, water depth is considered the single most


important factor for building damage (Elmer et al., 2010; Messner et al., 2007) and so is a

good characteristics to use.

The accuracy of such estimates can be difficult to validate, although comparisons can be made

to historical events and damage data. This is also reflective of the flood depth-damage

functions used by this study which are generally comprised of synthetic data and not directly

derived from analysis of historical flood events (Penning-Rowsell et al., 2010).

The depth damage functions are also based on national average data and so may underestimate

losses for London.

Additionally, it is reported that flood incidents can be accompanied by significant emergency

costs such as additional expenditure due to increased demand for police, fire and ambulance

services. It has been estimated that these are equivalent to 10.7% of property damages

(Penning-Rowsell et al., 2010). The above estimates reflect direct costs to residential property

only, but total costs including indirect effects to the wider region and economy, could be

substantially higher (e.g. see Crawford-Brown et al., 2012 for further discussion of indirect

losses).

Likewise, flood events can result in environmental and social impacts which are not

considered here e.g. environmental and ecological impacts of degraded water quality, and

psychological effects caused by displacement and stress following a flood event.

In the study affected houses were established by overlying the flood depth maps onto the

spatial footprints of residential buildings. This provides a simple interpretation of the houses

which would be affected given the limited data on the specific architecture of buildings (e.g.

height above pavement, whether they have basements etc.). This was a consequence of

developing a methodology which could be applied to a large region rather than focusing on

specific properties and streets.

Finally, the study considers one of the main drivers for flood risk – precipitation (and the

impact of climate change on precipitation regimes). However, other drivers such as

urbanisation will also be important. The impact of residential developments in areas of high

flood risk will be an area investigated further in the second stage of the analysis via the ABM.


5 Conclusions

The report presents a framework for evaluating current and future risk of surface water flooding in

Greater London. The case study highlights how this risk is expected to change in the future under

projections of climate change. Daily event frequency could increase by 54% by the 2030s and by 60%

by the 2050s compared to the baseline period. Average annual damage to residential properties is

projected to increase by 50% and 80% by the 2030s and 2050s respectively, suggesting an increase in

severity of events when they do occur. Based on the daily event data, the expected annual damage

(EAD) is calculated as £103, £184 and £198 million/year for the baseline, 2030 high and 2050 high

climate change scenarios respectively.

The risk analysis provides a useful tool for decision making and risk management, and through the

proposed incorporation of this data in the ABM aims to support and justify future adaptation

strategies. These will be particularly important for Greater London where surface water flooding is

considered to be the most likely cause of flood events, and probably the greatest short-term climate

risk.

As previously discussed a particularly interesting aspect of flood management in England is the

public-private partnership on flood insurance between the UK government and the insurance industry

and the new flood insurance system Flood Re. The proposed system, which creates an insurance pool

for properties at high risk of flooding, is presented by government and industry as a roadmap to future

affordability and availability of flood insurance, with an anticipated run-time of 20 to 25 years (Defra

and ABI 2013). Flood Re was proposed by government in summer 2013, with the political debate

about the system and its mechanisms continuing. The current aim is to finalise and implement the new

scheme by mid-2015, and as such the analysis of the scheme through the ABM proposed here is

highly relevant for informing the MSP.

The future link to the ABM will allow an estimate of costs and benefits of adaptation strategies, and

their role in risk reduction, to be investigated. In addition, ongoing discussions with key stakeholders

have highlighted specific questions and issues of interest. Through the ABM the following issues can

be investigated:

The resilience of the new Flood Re scheme to increases in surface water flood risks

How the design and coverage of the scheme can affect flood risk. For example how will the

inclusion/exclusion of post 2009 properties impact on the Flood Re scheme

How Flood Re could be used to enhance resilience. For example including clauses within the

scheme whereby homeowners need to demonstrate resilient repairs or their excess increases or they

are potentially excluded from Flood Re

Issues of affordability and availability of insurance at the end of the 25-year transition period

that Flood Re is supposed to operate for.

Modelling the transition from Flood Re to risk based pricing and mechanisms for this. For example, a

gradual removal of properties over time, staggered removal of properties based on council tax bands, or

a gradual increase in the price of premiums in Flood Re


The above issues and questions are all highly relevant aspects for the ongoing regulatory and political

approval process for Flood Re, which have until now not received sufficient attention due to lack of

data or analysis. Our work addresses this gap and our findings are expected to provide important input

to the current discussion about the design and operation of Flood Re, particularly with regards to

incentivising flood risk reduction measures.


6 References

Adaptation Sub-Committee, (2012) Climate Change - Is the UK preparing for flooding and water

scarcity? Committee on Climate Change, London, p. 102.

Atreya, A., Ferreira, S. (2014) Seeing is Believing? Evidence from Property Prices in Inundated

Areas. Risk Analysis, n/a-n/a.

Bevan, K., Hall, J., (2014) Applied Uncertainty Analysis for Flood Risk Management. Imperial

College Press, London, p. 684.

Brouwers, L., Boman, M., (2011) A computational agent model of flood management strategies, in: do

Prado, H.A., Luiz, A.J.B., Filho, H.C. (Eds.), Computational Methods for Agricultural Research:

Advances and Applications. IGI Global, Hershey, PA, USA, pp. 296–307.

Coulthard, T.J., Ramirez, J., Fowler, H.J., Glenis, V. (2012) Using the UKCP09 probabilistic

scenarios to model the amplified impact of climate change on river basin sediment yield. Hydrol.

Earth Syst. Sci. Discuss. 9, 8799-8840.

Crawford-Brown, D., Syddall, M., Guan, D., Li, J., Jenkins, K., Hall, J., Beaven, R., Thoung, C.,

Anger-Kraavi, A. (2012) Exploring the Vulnerability of London’s Economy to Climate Change

Impacts. Environment and Planning A Submitted.

Dawson, R.J., Speight, L., Hall, J.W., Djordjevi, S., Savic, D., Leandro, J. (2008) Attribution of flood

risk in urban areas. Journal of Hydroinformatics 10.4, 275-288.

Defra, (2008a) Consultation on policy options for promoting property-level flood protection and

resilience, in: Department for Environment, F.a.R.A. (Ed.). Department for Environment, Food and

Rural Affairs, London, p. 64.

Defra, (2008b) Resilience grants pilot projects, in: Department for Environment, F.a.R.A. (Ed.).

Department for Environment, Food and Rural Affairs, London, p. 15.

Defra, (2011) Commencement of the Flood and Water Management Act 2010, Schedule 3 for

Sustainable Drainage: Impact Assessment. Defra, London, p. 42.

Defra and ABI (2013), Flood Re Proposal: Memorandum of Understanding. London:

Department for Environment, Food and Rural Affairs, and the Association of British Insurers.

Available at:

https://consult.defra.gov.uk/flooding/floodinsurance/supporting_documents/20130626%20Fl

ood%20Insurance%20MOU%20June%202013%20unprotected.pdf [Accessed 8 July 2013].

Ellis, J.B., Viavattene, C. (2014) Sustainable Urban Drainage System Modeling for managing Urban

surface water Flood Risk. Clean - Soil, Air, Water 42, 153-159.

Elmer, F., Thieken, A.H., Pech, I., Kreibich, H. (2010) Influence of flood frequency on residential

building losses. Nat. Hazards Earth Syst. Sci., 10, 2145-2159.

Environment Agency, (2007) Science Report – A review of the cost-benefit of undertaking SUDS

retrofit in urban areas. Environment Agency, Bristol, p. 33.


Environment Agency, (2009) Flooding in England:A National Assessment of Flood Risk, in: Agency,

E. (Ed.). Environment Agency, Bristol, p. 36.

Environment Agency, (2010) Case study – Surface water flooding in London, July 2007. Environment

Agency, Avaliable at: http://www.environment-

agency.gov.uk/research/library/publications/115643.aspx.

European Water Association, (2009) EWA Expert Meeting on Pluvial Flooding in Europe, Final

Report. EWA, Brussels, Belgium.

Falconer, R.H., Cobby, D., Smyth, P., Astle, G., Dent, J., Golding, B. (2009) Pluvial flooding: new

approaches in flood warning, mapping and risk management. Journal of Flood Risk Management 2,

198-208.

Fowler, H.J., Ekström, M. (2009) Multi-model ensemble estimates of climate change impacts on UK

seasonal precipitation extremes. International journal of climatology 29, 385-416.

GLA, (2011) The London Plan: Spatial Development Strategy for Greater London. Greater London

Authority, London.

Greater London Authority, (2009) London Regional Flood Risk Appraisal, in: Plan, T.M.L. (Ed.).

GLA, London, p. 62.

Greater London Authority, (2011) The London Climate Change Adaptation Strategy. Greater London

Authority, London, p. 125.

Hall, J., Sayers, P., Dawson, R. (2005) National-scale assessment of current and future flood risk in

England and Wales. Natural Hazards 36, 147-164.

Hall, J.W., Sayers, P.B., Walkden, M.J.A., Panzeri, M. (2006) Impacts of climate change on coastal

flood risk in England and Wales: 2030–2100. Philosophical Transactions of the Royal Society A:

Mathematical, Physical and Engineering Sciences 364, 1027-1049.

Harries, T. (2008) Feeling secure or being secure? Why it can seem better not to protect yourself

against a natural hazard. Health, risk and society 10, 479-490.

Harries, T. (2012) The anticipated emotional consequences of adaptive behaviour - impacts on the

take-up of household flood-protection measures. Environment and Planning A 44, 649-668.

HR Wallingford, (2012) Development of spatial indicators to monitor changes in exposure and

vulnerability to flooding and the uptake of adaptation actions to manage flood risk in England: Results

2012. HR Wallingford Ltd, Oxfordshire, p. 56.

IPCC, (2013) Summary for Policymakers. In: Climate Change 2013: The Physical Science Basis.

Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on

Climate Change in: Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A.

Nauels, Y. Xia, V. Bex and P.M. Midgley. (Ed.). Cambridge University Press, Cambridge, United

Kingdom and New York, NY, USA., p. 29.

http://www.environment-agency.gov.uk/research/library/publications/115643.aspx

http://www.environment-agency.gov.uk/research/library/publications/115643.aspx


Jones, P.D., Kilsby, C.G., Harpham, C., Glenis, V., Burton, A., (2009) UK Climate Projections

science report: Projections of future daily climate for the UK from the Weather Generator, University

of Newcastle, UK.

Kilsby, C., Jones, P., Harpham, C., Glenis, V., Burton, A., (2011) Spatial Urban weather Generator for

Future Climates, ARCADIA Task 3 Report. Avaliable at: http://www.arcc-cn.org.uk/wp-

content/pdfs/ARCADIA-7.pdf, p. 8.

Lamond, J., Proverbs, D. (2006) Does the price impact of flooding fade away? Structural Survey 24,

363-377.

Lamond, J., Proverbs, D., Hammond, F., (2009) Flooding and Property Values, in: Brown, S. (Ed.),

Findings in Built and Rural Environments (FiBRE). Royal Institution of Chartered Surveyors, London.

Maraun, D., Osborn, T.J., Gillett, N.P. (2008) United Kingdom daily precipitation intensity: improved

early data, error estimates and an update from 2000 to 2006. International journal of climatology 28,

833-842.

Messner, F., Penning-Rowsell, E., Green, C., Meyer, V., Tunstall, S., van der Veen, A., (2007)

Evaluating flood damages: guidance and recommendations on principles and methods. HR

Wallingford, UK p. 189.

Murphy, J.M., Sexton, D.M.H., Jenkins, G.J., Boorman, P.M., Booth, B.B.B., Brown, C.C., Clark,

R.T., Collins, M., Harris, G.R., Kendon, E.J., Betts, R.A., Brown, S.J., Howard, T.P., Humphrey,

K.A., McCarthy, M.P., McDonald, R.E., Stephens, A., Wallace, C., Warren, R., Wilby, R., Wood,

R.A., (2009) UK Climate Projections Science Report: Climate change projections, Met Office Hadley

Centre, Exeter, UK, p. 190.

ONS, (2001) Census: Standard Area Statistics (England and Wales), ESRC/JISC Census Programme,

Census Dissemination Unit. Mimas, University of Manchester.

Penning-Rowsell, E., Viavattene, C., Pardoe, J., Chatterton, J., Parker, D., Morris, J., (2010) The

Benefits of Flood and Coastal Risk Management: A Handbook of Assessment Techniques. Flood

Hazard Research Centre, London, p. 98.

Ramsbottom, D., Sayers, P., Panzeri, M., (2012).Climate Change Risk Assessment for the Floods and

Coastal Erosion Sector. UK Climate Change Risk Assessment. Defra, London.

Sanderson, M., (2010) Changes in the frequency of extreme rainfall events for selected towns and

cities, Report for Ofwat. Met Office, Exeter, p. 38.

Thurston, N., Finlinson, B., Breakspear, R., Williams, N., Shaw, J., Chatterton, J., (2008) Developing

the evidence base for flood resilience, FD2607/TR. Defra, London.

Ulbinaitė, A., Kučinskienė, M., Le Moullec, Y. (2011) Conceptualising and Simulating Insurance

Consumer Behaviour: an Agent-Based-Model Approach. International Journal of Modeling and

Optimization 1, 250-257.

Ulbinaitė, A., Le Moullec, Y. (2010) Towards an ABM-based framework for investigating consumer

behaviour in the insurance industry. ekonomika 89, 95-110.

http://www.arcc-cn.org.uk/wp-content/pdfs/ARCADIA-7.pdf

http://www.arcc-cn.org.uk/wp-content/pdfs/ARCADIA-7.pdf


Ward, P.J., de Moel, H., Aerts, J.C.J.H. (2011) How are flood risk estimates affected by the choice of

return-periods? Nat. Hazards Earth Syst. Sci. 11, 3181-3195.

Zhang, H., Putra, H.C., Andrews, C.J., (2013) Modeling Real Estate Market Responses to Climate

Change in The Coastal Zone, The Computational Social Science Society of the Americas, Santa Fe,

USA, p. 18.

ENHANCE - Environmental Change Institute€¦ · A particularly interesting aspect of flood management in England is the public-private partnership on flood insurance between the

Documents